Acoustic_2019_Explore3
Sean Dimoff
October 9, 2019
Exploratory Plots for 2017-2018 Acoustic/Fish Data
Purpose To explore the Acoustic data gathered in 2017 and 2018 to expose important trends between sites, diurnal patterns, fish abundance, lunar phase, and coral reef acoustics.
Validations
Combined Model All variables are matched to the files that were used for Fish call counts (3:00, 9:00, 15:00, 21:00)
Confidence Intervals
Distributions
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.
Scatterplots
Scatterplots SPL
Plotting explanatory (Knocks, Calls, Herbivory, Snaps) against response variables (MF and HF) SPL 
Patterns Breakdown SPL and ACI
Patterns in HF and Snaps at Site Level - SPL
Patterns in Tot_Knocks and Hour at Site and Hour Level - SPL
Breaking down the relationship between total knocks and MF to the site and hour level 
Regressions
Biotic Regressions - SPL
Running basic regressions linking the explanatory to the response at their lowest levels and combined to see how different sites/ hours change the regression - SPL
Linear Model outputs below each
##
## Call:
## lm(formula = SPL_HF ~ Snaps, data = Snap.HF17)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.8309 -1.9842 0.2062 1.8451 13.3944
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.053e+02 6.541e-01 160.99 <2e-16 ***
## Snaps 7.227e-03 4.475e-04 16.15 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.807 on 10163 degrees of freedom
## Multiple R-squared: 0.02502, Adjusted R-squared: 0.02493
## F-statistic: 260.8 on 1 and 10163 DF, p-value: < 2.2e-162017 Snap data, snaps significant.
##
## Call:
## lm(formula = SPL_HF ~ Snaps, data = Snap.HF18)
##
## Residuals:
## Min 1Q Median 3Q Max
## -9.4682 -1.9696 0.0058 2.4042 30.2074
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 8.617e+01 8.999e-01 95.75 <2e-16 ***
## Snaps 2.269e-02 6.168e-04 36.78 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.142 on 5823 degrees of freedom
## Multiple R-squared: 0.1886, Adjusted R-squared: 0.1884
## F-statistic: 1353 on 1 and 5823 DF, p-value: < 2.2e-162018 Snap data with outliers removed. Snaps significant.
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = AC.DF1)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.248 -2.267 -0.871 1.597 19.211
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.047e+02 3.888e-01 269.163 < 2e-16 ***
## Tot_Knocks 1.744e-02 4.465e-03 3.906 0.000129 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.519 on 198 degrees of freedom
## Multiple R-squared: 0.07154, Adjusted R-squared: 0.06685
## F-statistic: 15.26 on 1 and 198 DF, p-value: 0.00012872017-2018 data w/200 samples. 1st plot splits by site and second by hour to show any patterns before I break them down individually.
Breakdown by Site
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = s5)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.4846 -2.3049 0.1011 1.9482 6.0310
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.065e+02 1.106e+00 96.290 <2e-16 ***
## Tot_Knocks 5.551e-04 8.256e-03 0.067 0.947
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.034 on 38 degrees of freedom
## Multiple R-squared: 0.0001189, Adjusted R-squared: -0.02619
## F-statistic: 0.00452 on 1 and 38 DF, p-value: 0.9467Site 5, knocks not significant.
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = s35)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10.1201 -3.6626 0.4059 4.2686 9.1758
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 105.01636 1.19662 87.761 <2e-16 ***
## Tot_Knocks 0.03231 0.01218 2.653 0.0116 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.804 on 38 degrees of freedom
## Multiple R-squared: 0.1563, Adjusted R-squared: 0.1341
## F-statistic: 7.039 on 1 and 38 DF, p-value: 0.01157Site 35, knocks significant.
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = s8)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.5526 -1.5016 0.6098 1.8588 6.6098
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 105.497101 0.700474 150.61 <2e-16 ***
## Tot_Knocks -0.006653 0.009929 -0.67 0.507
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.727 on 38 degrees of freedom
## Multiple R-squared: 0.01168, Adjusted R-squared: -0.01433
## F-statistic: 0.449 on 1 and 38 DF, p-value: 0.5068Site 8, knocks not significant. Negative relationship… thats interesting.
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = s40)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.2090 -0.9792 -0.3831 0.7009 4.7409
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.041e+02 4.407e-01 236.176 <2e-16 ***
## Tot_Knocks 6.514e-03 8.094e-03 0.805 0.426
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.554 on 38 degrees of freedom
## Multiple R-squared: 0.01676, Adjusted R-squared: -0.009116
## F-statistic: 0.6477 on 1 and 38 DF, p-value: 0.4259Site 40, knocks not significant.
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = s32)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.0442 -1.9728 -0.7078 0.0613 18.4340
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 103.79253 0.92602 112.084 <2e-16 ***
## Tot_Knocks 0.04784 0.01903 2.514 0.0163 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.783 on 38 degrees of freedom
## Multiple R-squared: 0.1426, Adjusted R-squared: 0.12
## F-statistic: 6.321 on 1 and 38 DF, p-value: 0.01629Site 32, knocks significant.
Breakdown by Hour
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = h3)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.8821 -2.3813 -0.5447 2.0264 6.8553
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.043e+02 7.055e-01 147.893 <2e-16 ***
## Tot_Knocks 5.296e-03 7.304e-03 0.725 0.472
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.121 on 48 degrees of freedom
## Multiple R-squared: 0.01083, Adjusted R-squared: -0.009773
## F-statistic: 0.5258 on 1 and 48 DF, p-value: 0.47193AM, knocks not significant
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = h9)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.3144 -1.6662 -0.4952 0.7818 8.0555
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 102.90908 0.61924 166.186 < 2e-16 ***
## Tot_Knocks 0.05274 0.00653 8.076 1.69e-10 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.703 on 48 degrees of freedom
## Multiple R-squared: 0.5761, Adjusted R-squared: 0.5672
## F-statistic: 65.22 on 1 and 48 DF, p-value: 1.69e-109AM, knocks significant
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = h15)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.0816 -2.0625 -0.9435 1.2227 7.1127
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 105.697228 0.669185 157.949 <2e-16 ***
## Tot_Knocks -0.006816 0.011700 -0.583 0.563
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.128 on 48 degrees of freedom
## Multiple R-squared: 0.007021, Adjusted R-squared: -0.01367
## F-statistic: 0.3394 on 1 and 48 DF, p-value: 0.56293PM, knocks not significant
##
## Call:
## lm(formula = SPL_Midrange ~ Tot_Knocks, data = h21)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.987 -2.505 -0.915 1.457 18.595
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.060e+02 9.217e-01 114.979 <2e-16 ***
## Tot_Knocks 4.355e-03 9.860e-03 0.442 0.661
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.91 on 48 degrees of freedom
## Multiple R-squared: 0.004048, Adjusted R-squared: -0.0167
## F-statistic: 0.1951 on 1 and 48 DF, p-value: 0.66079PM, knocks not significant.
Summary Knocks significantly explained SPLMF at sites 35 and 32 and at 9AM.
Abiotic Regressions (Wind) -SPL
Running basic regressions linking the wind to SPL at both HF and MF to see if wind speed is significantly affecting the sound
## Warning: Removed 1518 rows containing non-finite values (stat_smooth).## Warning: Removed 1518 rows containing missing values (geom_point).
## Warning: Removed 1520 rows containing non-finite values (stat_smooth).## Warning: Removed 1520 rows containing missing values (geom_point).
Wind doesn’t seem to impact SPL HF or MF in any particular direction. Although the wind range seems really small.
Time-Series Acoustics
Acoustics Breakdown All acoustic metrics (SPL and ACI) are broken down into 3 frequency bands: Broadband (All frequencies), High Frequency (Frequencies between 1 kHz - 22 kHz), and Mid Frequency (Frequencies between 160 Hz and 1 kHz)
Note 2017 had a 10 minute duty cycle with 5 minutes recording while 2018 had a 15 minute duty cycle with 5 minutes recording, so the number of files averages differs between years
Frequency-Band Breakdowns
High Frequency
Plots of high frequency patterns, notice diurnal patterns with highest SPL at night and lowest during the day (this is shown in the literature), also notice the clear splits by site.
Notice, site 35 seems to have switched position between 2017 and 2018 but all of the other sites seem to be staying more or less in the same spot
- I wonder if this has to do with macroalgal cover changes between 2017 and 2018
Mid Frequency
Plots of mid frequency patterns, notice opposite diurnal patterns with highest SPL during the day and lowest at night, also notice the clear splits by site.
Also, notice that 35 does a similar switch in Mid-frequency, going from the bottom in 2017 to the top in 2018
Combo Plots
Models
Preliminary Models Looking into the relationships between biogenic sounds (Knocks/Calls and Snaps) and their frequency spectra (MF SPL/HF SPL) respectively.
Model 1
Looking at Total Knocks only SPL MF ~ Tot_Knocks
#model 1 looking at Total Knocks only
gfit1 <- glm(SPL_Midrange ~ Tot_Knocks, data = AC.DF1, family = Gamma)
summary(gfit1)##
## Call:
## glm(formula = SPL_Midrange ~ Tot_Knocks, family = Gamma, data = AC.DF1)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.068872 -0.021698 -0.008332 0.015069 0.171982
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.553e-03 3.461e-05 276.019 < 2e-16 ***
## Tot_Knocks -1.534e-06 3.914e-07 -3.918 0.000123 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for Gamma family taken to be 0.001098715)
##
## Null deviance: 0.22855 on 199 degrees of freedom
## Residual deviance: 0.21186 on 198 degrees of freedom
## AIC: 1068.1
##
## Number of Fisher Scoring iterations: 3par(mfrow = c(2,2))
plot(gfit1)
summary.glm(gfit1)$coefficients## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.553059e-03 3.461011e-05 276.019343 5.079866e-258
## Tot_Knocks -1.533644e-06 3.913970e-07 -3.918385 1.227169e-04Model 2
Looking at Total Knocks and Number of Long Calls SPL MF ~ Tot_Knocks + Num_L_Calls
#model 1 looking at Total Knocks only
gfit2 <- glm(SPL_Midrange ~ Tot_Knocks + Num_L_calls, data = AC.DF1, family = Gamma)
summary(gfit2)##
## Call:
## glm(formula = SPL_Midrange ~ Tot_Knocks + Num_L_calls, family = Gamma,
## data = AC.DF1)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.068874 -0.021712 -0.008334 0.015083 0.171968
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.553e-03 4.005e-05 238.518 < 2e-16 ***
## Tot_Knocks -1.533e-06 3.946e-07 -3.886 0.000139 ***
## Num_L_calls 2.677e-08 3.231e-06 0.008 0.993399
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for Gamma family taken to be 0.001104281)
##
## Null deviance: 0.22855 on 199 degrees of freedom
## Residual deviance: 0.21186 on 197 degrees of freedom
## AIC: 1070.1
##
## Number of Fisher Scoring iterations: 3par(mfrow = c(2,2))
plot(gfit2)
summary.glm(gfit2)$coefficients## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.552893e-03 4.005097e-05 238.518388890 1.738362e-244
## Tot_Knocks -1.533301e-06 3.945605e-07 -3.886099544 1.391033e-04
## Num_L_calls 2.676589e-08 3.230984e-06 0.008284128 9.933987e-01Model 3
Looking at Total Knocks/Number of long calls/Herbivory SPL MF ~ Tot_Knocks + Num_L_Calls + Num_Herbivory
#model 1 looking at Total Knocks only
gfit3 <- glm(SPL_Midrange ~ Tot_Knocks + Num_L_calls + Num_Herbivory, data = AC.DF1, family = Gamma)
summary(gfit3)##
## Call:
## glm(formula = SPL_Midrange ~ Tot_Knocks + Num_L_calls + Num_Herbivory,
## family = Gamma, data = AC.DF1)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -0.067662 -0.021756 -0.007807 0.015801 0.173266
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.565e-03 4.063e-05 235.409 < 2e-16 ***
## Tot_Knocks -1.539e-06 3.932e-07 -3.915 0.000125 ***
## Num_L_calls 3.309e-07 3.225e-06 0.103 0.918400
## Num_Herbivory -3.975e-06 2.555e-06 -1.556 0.121409
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for Gamma family taken to be 0.00109656)
##
## Null deviance: 0.22855 on 199 degrees of freedom
## Residual deviance: 0.20923 on 196 degrees of freedom
## AIC: 1069.6
##
## Number of Fisher Scoring iterations: 3par(mfrow = c(2,2))
plot(gfit3)
summary.glm(gfit3)$coefficients## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 9.564541e-03 4.062942e-05 235.4092305 2.341236e-242
## Tot_Knocks -1.539367e-06 3.932317e-07 -3.9146563 1.248754e-04
## Num_L_calls 3.308596e-07 3.225337e-06 0.1025814 9.184001e-01
## Num_Herbivory -3.975108e-06 2.555300e-06 -1.5556324 1.214087e-01Model 4 - HF SPL and Snaps
Looking at Snaps and their effect on the HF SPL SPL HF ~ Snaps Distributions look normal so this is a linear model
fit4 <- lm(SPL_HF ~ Snaps, data = AC.DF1)
summary(fit4)##
## Call:
## lm(formula = SPL_HF ~ Snaps, data = AC.DF1)
##
## Residuals:
## Min 1Q Median 3Q Max
## -7.4772 -2.6200 -0.4764 2.6614 8.3553
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 91.690030 5.741405 15.970 < 2e-16 ***
## Snaps 0.017654 0.003924 4.499 1.16e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.549 on 198 degrees of freedom
## Multiple R-squared: 0.09275, Adjusted R-squared: 0.08817
## F-statistic: 20.24 on 1 and 198 DF, p-value: 1.162e-05par(mfrow = c(2,2))
plot(fit4)
summary(fit4)$coefficients## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 91.69002981 5.741405123 15.969963 1.943186e-37
## Snaps 0.01765414 0.003923967 4.499054 1.162338e-05